the greenscape shapes surfing of resource waves in a large migratory herbivore · 2020-04-07 ·...

LETTER The greenscape shapes surfing of resource waves in a large

migratory herbivore

Ellen O. Aikens,1,3*

Matthew J. Kauffman,2,3

Jerod A. Merkle,1

Samantha P. H. Dwinnell,1

Gary L. Fralick4 and

Kevin L. Monteith1,3,5

Abstract

The Green Wave Hypothesis posits that herbivore migration manifests in response to waves ofspring green-up (i.e. green-wave surfing). Nonetheless, empirical support for the Green WaveHypothesis is mixed, and a framework for understanding variation in surfing is lacking. In a pop-ulation of migratory mule deer (Odocoileus hemionus), 31% surfed plant phenology in spring aswell as a theoretically perfect surfer, and 98% surfed better than random. Green-wave surfing var-ied among individuals and was unrelated to age or energetic state. Instead, the greenscape, whichwe define as the order, rate and duration of green-up along migratory routes, was the primary fac-tor influencing surfing. Our results indicate that migratory routes are more than a link betweenseasonal ranges, and they provide an important, but often overlooked, foraging habitat. In addi-tion, the spatiotemporal configuration of forage resources that propagate along migratory routesshape animal movement and presumably, energy gains during migration.

Keywords

Forage maturation hypothesis, green wave hypothesis, migration, mule deer, normalised differencevegetation index, Odocoileus hemionus, phenology, resource landscape, ungulate, Wyoming.

Ecology Letters (2017) 20: 741–750

INTRODUCTION

The resource landscape – i.e. the spatiotemporal configurationof resources – strongly influences behavioural strategies ofmobile organisms. Resource landscapes where resources aretemporally predictable and spatially homogeneous favour resi-dent (i.e. home range) movement strategies (Mueller & Fagan2008). Resource landscapes also can be characterised byresource waves, where natural gradients (e.g. elevation) createpulses of resources that propagate across space and time. Classicexamples of resource waves include the onset of spring green-upand emergence of insects across latitudinal gradients (Hodkin-son 2005; Moser et al. 2010). Resource waves should favourmovement strategies where animals match their movements withresources as they propagate across the landscape (Armstronget al. 2016). From grizzly bears (Ursus arctos) pursuing sockeyesalmon (Oncorhynchus nerka) across spawning sites (Schindleret al. 2013), to surf scoters (Melanitta perspicillata) pursuingwaves of spawning herring (Clupea pallasii; Lok et al. 2011),consumers enhance foraging by tracking resource waves.For herbivores, plant phenology is a key factor that shapes

the resource landscape. Herbivores balance forage quality andquantity by selecting for vegetation at intermediate biomass(hereafter green-up) because net energy intake is mediated bya trade-off between intake rate and digestibility (i.e. The For-age Maturation Hypothesis; Fryxell 1991). Because plantgrowth is delayed at high elevations and latitudes in spring,green-up moves as a resource wave, often across vast areas.

The Green Wave Hypothesis (GWH) provides a conceptualframework to understand how the movements of herbivoresduring migration are timed to respond to resource waves. Inaccordance with the GWH, large-scale movements of herbi-vores should track waves of green-up that propagate acrossthe landscape (Drent et al. 1978). For example, barnacle geese(Branta leucopsis) appear to track or ‘surf’ waves of plantgreen-up across a latitudinal gradient during migration (Shari-atinajafabadi et al. 2014).Viewing migration as a movement strategy that is in part

driven by resource waves challenges traditional conceptualisa-tions of migration. Migration is commonly viewed as themovement between two distinct seasonal ranges, which allowsmigrants to exploit resource abundances in one seasonal rangewhile avoiding resource deficits in the other (Alerstam et al.2003). Although this view of migration is generally supported(e.g. Shaffer et al. 2006), it ignores how resources acquired enroute influence movement and energy gain during migration(Avgar et al. 2013). Migration has even been defined as move-ment that is ‘undistracted’ by surrounding resources (Dingle2014). The GWH suggests an alternative view, where migra-tory movements themselves prolong exposure to high-qualityresources and serve to enhance energy gain.Accumulating evidence supports the notion that some

migratory ungulates surf green waves of forage during migra-tion (Avgar et al. 2013). The migration of wildebeest (Con-nochaetes taurinus) in the Serengeti appears to result fromindividuals seeking habitat patches flush with new vegetation

1Wyoming Cooperative Fish and Wildlife Research Unit, Department of

Zoology and Physiology, University of Wyoming, Laramie, WY 82071, USA2U.S. Geological Survey, Wyoming Cooperative Fish and Wildlife Research

Unit, Department of Zoology and Physiology, University of Wyoming,

Laramie, WY 82071, USA

3Program in Ecology, University of Wyoming, Laramie, WY 82071, USA4Wyoming Game and Fish Department, Thayne, WY 83127, USA5Haub School of Environment and Natural Resources, University of Wyoming,

Laramie, WY 82072, USA

*Correspondence: E-mail: [email protected]

© 2017 John Wiley & Sons Ltd/CNRS

Ecology Letters, (2017) 20: 741–750 doi: 10.1111/ele.12772

(Boone et al. 2006). During spring migration, mule deer (Odo-coileus hemionus) in the Rocky Mountains spend prolongedamounts of time (95% of the migratory period) at stopoversites, and use of stopover sites is synchronised with green-up(Sawyer & Kauffman 2011). Moreover, red deer (Cervus ela-phus) that experienced more heterogeneous plant phenology(akin to experiencing a longer green wave) grow larger thanthose that experience less variable phenology (Albon & Lang-vatn 1992; Mysterud et al. 2001), which suggests a fitness ben-efit associated with green-wave surfing.Despite indirect support for the GWH and its strong concep-

tual underpinnings, the first explicit test of the GWH in anungulate revealed that red deer do not surf (Bischof et al. 2012).Instead, red deer ‘jump’ the green wave, moving rapidly fromwinter range to summer range, without regard to the progres-sion of the green wave across the landscape. Nevertheless,ungulates should have the capability to surf because several spe-cies select habitat patches at peak green-up (Merkle et al. 2016),and African ungulates appear to track gradients of plantgrowth and precipitation during their seasonal movements(Fryxell et al. 2004; Holdo et al. 2009; Bartlam-Brooks et al.2013; Bohrer et al. 2014). These findings highlight an impor-tant, yet unresolved, question of whether green-wave surfing isa common behaviour among migrating ungulates. Answeringthis question will help shape our ecological understanding ofmigration in ungulates, especially as to whether the key benefitof migration is simply access to seasonal ranges, or if the migra-tory route itself also provides a foraging benefit.Despite ongoing debate regarding the existence of green-

wave surfing, factors that facilitate or constrain surfing arealmost completely unexplored. The way that the green waveprogresses across the landscape – a concept we refer to as the‘greenscape’ – could be an important factor influencing surf-ing. For example, rapid green-up is thought to influencedemography by reducing the amount of time that high-qualityforage is available (Middleton et al. 2013). Furthermore, rapidgreen-up has been linked to declines in recruitment of youngfor several ungulate species (Pettorelli et al. 2007; Monteithet al. 2015). Such demographic consequences of rapid green-up may manifest, in part, by constraining the ability of herbi-vores to optimally surf phenological gradients. To conceptu-alise this idea, we propose the Greenscape Hypothesis,wherein the speed, duration and order of green-up along amigratory route influence the ability of animals to surf greenwaves of forage. Accordingly, we predicted that migratoryroutes with resource waves that green-up gradually, last for along time, and progress consecutively from one seasonal rangeto the other, should enhance surfing.Several other factors could affect green-wave surfing. First,

surfing may involve behavioural trade-offs (e.g. parental care,socialisation, risk avoidance) such that the reproductive ornutritional state of an animal influences behaviour duringmigration. State-dependent foraging exists in a wide range ofanimal taxa (e.g. Hutchings et al. 2001; Heithaus et al. 2007),where animals in poor condition prioritise foraging to enhanceenergy gain over other behaviours (e.g. vigilance; Mcnamara &Houston 1992). Green-wave surfing is fundamentally a foragingbehaviour and thus, we extend the State-Dependent Hypothe-sis, and predict that individuals in poor condition, or those that

anticipate high energetic costs (e.g. lactation), should prioritisegreen-wave surfing to maximise energy gains.Memory and learning also may influence green-wave surf-

ing. Resource waves in natural environments are inherentlynoisy (Guttal & Couzin 2010), and taxis along noisy environ-mental gradients is inefficient (Grunbaum 1998). Efficientsurfing should be contingent upon knowing when and whereto move to access high-quality forage, which can be enhancedby social interactions (Berdahl et al. 2013; Merkle et al.2015), learning (Mueller et al. 2013) and memory (Merkleet al. 2014). Spatial memory influences foraging efficiencyacross a wide range of taxa (Fagan et al. 2013), includingungulates (Merkle et al. 2014), and older animals have beenshown to migrate more efficiently than inexperienced individu-als (Mueller et al. 2013). We propose the Learning Hypothe-sis, and posit that efficient green-wave surfing must be learnedand thus surfing should improve with experience. We predictthat older, more experienced individuals should be more profi-cient at surfing than young, inexperienced individuals.We tested the GWH and assessed empirical support for

hypotheses that explain individual variability in green-wavesurfing using GPS collar data from migratory mule deer inWyoming, USA. We first tested the prediction that deershould use foraging patches along their migratory route whenthey are at peak green-up, providing a quantitative test of theGWH. Second, we quantified how surfing differed among3 years (with variable precipitation) and across short-, moder-ate- and long-duration migrations. Third, we examined empir-ical support for the Greenscape and State-DependentHypotheses, and a modified version of the Learning Hypothe-sis to explain individual variation in surfing. Our findings pro-vide empirical support for the GWH and the GreenscapeHypothesis – suggesting that access to plant green-up alongthe migratory route is a key foraging benefit of migration,and that certain phenological characteristics of migratoryroutes promote green-wave surfing.

METHODS

Study area

Western Wyoming, USA (42°250N, 110°420W), is a semi-aridregion in the Rocky Mountains characterised by long, cold win-ters and warm summers (see Appendix S1 in Supporting Infor-mation). Low elevations receive approximately 18 cm ofprecipitation per year (Big Piney, WY weather station,2080 m), and high elevations receive approximately 5 times thatamount (Spring Creek Divide SNOTEL site, 2750 m). Low ele-vations (~ 1800–2300 m) are dominated by sagebrush (Arteme-sia spp.), with sparse mountain shrub communities and willowcomplexes. Middle elevations (~ 2300–2750 m) include aspen(Populus tremuloides), sagebrush, mixed-mountain shrubs andconifers, and high elevations (> 2750 m) contain pine and firspecies and tall forb communities.

Animal capture

In March and December 2013–2015, we captured and fit GPScollars on 99 adult (> 1-year-old; age range = 2–12) female


742 E. O. Aikens et al. Letter

mule deer from two winter ranges via helicopter netgun (Bar-rett et al. 1982). We extracted one incisiform canine to esti-mate age by cementum annuli (Bleich et al. 2003). EachMarch, we recaptured individuals to quantify nutritional con-dition (ingesta-free body fat), pregnancy, and foetal number(Appendix S2). All animal capture and handling protocolswere approved by an independent Institutional Animal Careand Use Committee at the University of Wyoming (Protocol#20151204KM00135–01).

GPS telemetry data

We programmed GPS collars (Advanced Telemetry SystemsInc, Isanti, MN, USA and Telonics Inc, Mesa, AZ, USA) tocollect fixes every 1–5 h. We quantified surfing during two timeperiods: the period of migration (defined by the start and end ofspring migration using GPS data) and the period of springgreen-up (defined as the time when green-up was availableacross the study area). We used Net Squared Displacement(NSD) to delineate the start and end of spring migration foreach deer annually (Fig. S3; Appendix S3; Bunnefeld et al.2011). To delineate the physical pathway that individuals usedduring migration, hereafter the migratory route, we first inter-polated points between fixes, using a correlated random walkmodel (Johnson et al. 2008). We then applied the Visvalingamline simplification algorithm (Harrower & Bloch 2006) to theinterpolated paths, which removed tortuosity (Appendix S4).We simplified the interpolated path so that areas heavily usedby deer (i.e. stopovers) were not oversampled in our calculationof the greenscape, because the greenscape is intended to quan-tify patterns of plant phenology irrespective of the movement ofindividuals. We used the simplified paths representing migra-tory routes solely to calculate the greenscape, whereas we calcu-lated green-wave surfing scores using GPS locations from deer.

Metrics of green-wave surfing

We indexed vegetation quality across space and time by esti-mating the Instantaneous Rate of Green-up (IRG), a metricderived from a time series of the Normalised Difference Vegeta-tion Index (NDVI; MOD09Q1; 250-m spatial resolution, 8-daytemporal resolution; Appendix S5). The NDVI was correlatedwith vegetation biomass in a similar study area (Garroutte et al.2016). To calculate IRG, we pre-processed NDVI data and fitdouble-logistic curves to time series following Bischof et al.(2012) and Merkle et al. (2016; Fig. 1a; Appendix S5). TheIRG is the first derivative of the fitted NDVI curve, whichresults in a curve that peaks when vegetation growth is mostrapid (i.e. intermediate biomass), hereafter peak green-up(Fig. 1a). We defined the period of spring green-up as the daterange when IRG peaked across our study area. We defined ourstudy area as the 95% minimum convex polygon of all GPSlocations of deer, and defined the start and end of spring green-up as the 0.02 and 0.98 quantiles of peak IRG dates within thestudy area (Appendix S6).To evaluate variation in green-wave surfing among individu-

als, we used two metrics of surfing: (1) the IRG experiencedduring spring green-up (Bischof et al. 2012); and (2) the days-from-peak IRG, hereafter Days-From-Peak, which was calcu-lated as the absolute difference (in days) between the date a

deer used a given location and date of peak IRG at that samelocation. We calculated both metrics for each GPS location,and then averaged by day to reduce potential bias from incon-sistent fix rates and, thus, unbalanced sample size. We calcu-lated surfing scores, using both IRG and Days-From-Peak,for each animal (n = 99) during spring green-up in 2013, 2014and 2015 (total deer-years = 191).Although large values of IRG and small values of Days-

From-Peak indicate that an animal matches its movementswith peaks in IRG, there is an important but subtle differencebetween the two metrics. The IRG is comprised of twoparameters, one that describes when peak green-up is reached,and another that controls the speed at which that peak isreached (the spring scale which corresponds to the reciprocalof the green-up rate; Bischof et al. 2012). Two locations onthe landscape can have the same date of peak green-up, butdifferent rates at which green-up occurred, which influencesthe shape of the IRG curve. If a deer visited these two loca-tions 5 days before peak green-up, it would receive differentIRG scores solely because of different rates of green-up(Fig. 1b–c). Therefore, we view IRG as a metric that reflectsboth animal behaviour and the environment, whereas Days-

Figure 1 Hypothetical illustration of annual change in forage biomass

(grey dashed line) estimated by a double-logistic curve fit to a scaled time

series of NDVI data, and forage quality (solid black line) represented as

the Instantaneous Rate of Green-up (IRG) or first derivative of the

NDVI curve for one patch along a migratory route (a). We quantified

green-wave surfing using mean IRG and Days-From-Peak (DFP) IRG,

which measures the absolute difference in days between deer use of a

patch and the date of peak IRG. An IRG value of 1 or a DFP value of 0

represents perfect tracking of green-up. Because IRG is a measure of both

environment and behaviour, equivalent tracking (e.g. DFP = 5) can

dramatically reduce IRG when green-up occurs rapidly (b) compared to

patches with more gradual green-up (c).


Letter Surfing green waves of forage 743

From-Peak reflects only behaviour. We explored how dura-tion of migration influenced green-wave surfing by examiningannual patterns in Days-From-Peak and IRG for migrationsthat were classified as short (< 0.33 quantile), mid (0.33–0.66quantile) and long (> 0.66 quantile) duration migrations.

Metrics of the greenscape

We calculated three metrics to characterise the greenscape:green-up rate, green-up order and green-up duration. We esti-mated the date of peak green-up and rate of green-up at 1-kmintervals along the migratory route. We used the reciprocal ofthe rate of green-up – the spring scale, which can be inter-preted as an index of the time to reach peak green-up – toavoid subsequent issues associated with skewed distributionsin green-up rate. We calculated average spring scale of the 1-km intervals, as our metric of green-up rate. To quantifygreen-up order, we calculated Spearman’s rank correlationbetween the physical order of each location along the migra-tory route (1-km interval) and the chronological order inwhich those locations reached peak green-up. A rank correla-tion of one represents a migratory route that greens up con-secutively from winter to summer range, whereas a rankcorrelation of 0 indicates random green-up order. We deter-mined the duration of green-up of a route by calculating thedifference between the mean Julian date of peak green-up atthe first and last location (buffered by 500 m) along themigratory route. We buffered the locations to avoid erroneousendpoints that may not be representative of the NDVI signa-tures around them.

Statistical analysis

We tested the key prediction of the GWH by comparing thedate that a deer occupied a habitat patch to the date of peakgreen-up of that habitat patch. If deer surf according to theGWH, we expect a strong, positive relationship with a slopeof one and an intercept of zero. We used linear regression toestimate the intercept and slope of the relationship betweendate of peak green-up and date of deer use, for each individ-ual during both the year’s spring green-up period and migra-tion period. We classified each regression as: (1) theoreticallyperfect surfing (i.e. 95% confidence intervals [CIs] of the slopeoverlapped one and 95% CIs of the intercept overlapped withzero); (2) surfing that was better than random (i.e. slope waspositive and 95% CIs did not overlap zero) or (3) not surfing(i.e. a negative slope or a slope with 95% CIs overlappingzero).To examine the factors influencing green-wave surfing

among individuals, we modelled IRG and Days-From-Peak asa function of predictor variables that corresponded to theLearning (age), Greenscape (rate, duration and order ofgreen-up) and State-Dependent (nutritional condition and foe-tal number) Hypotheses. Although learning is critical duringearly life stages (Wunderle 1991), we were unable to obtainfine-scale movement data for young-of-the-year deer. Wetherefore reframed this hypothesis as the Adult LearningHypothesis and evaluated the effect of increased experienceon surfing proficiency for adults. We transformed Days-From-

Peak using the natural log, and arcsine transformed green-uporder, because their distributions were skewed. We used aninformation-theoretic approach to assess the relative supportfor hypotheses (Burnham & Anderson 2002). We modelledIRG and Days-From-Peak using a linear mixed-effects modelthat included a random intercept to account for variabilityacross years. Using a two-step modelling procedure, we firstdeveloped a global model based on predictor variables thatcorresponded to each hypothesis and subsequently evaluatedall possible combinations of the predictor variables, withoutallowing collinear variables to enter the same model (Dohertyet al. 2012). Then, to improve the precision of coefficient esti-mates, we reparametrised top models (those within two DAICc

units from the best model) with random slopes, so that eachvariable in the model could vary by year (Bolker 2015). Weconsidered predictor variables within top models as significantif the 95% CIs of the coefficient estimate did not overlapzero.To visualise the effect of the greenscape across our study

area, we mapped predicted IRG from deer monitored acrossthe 3-year study period (n = 32). We derived predicted IRGbased on fixed-effect coefficients estimated from the top rank-ing IRG model based solely on greenscape metrics. Using apermutation test, we compared the greenscape across yearsand across routes to determine: (1) if the greenscape was dif-ferent from the null hypothesis of no difference in the green-scape among years; and (2) if individual routes maintainedconsistently high- or low-quality greenscapes (i.e. routes withgreenscapes that were easier or more difficult to surf) againstthe null hypothesis that there was no difference in qualityamong routes. We calculated P-values for the permutationtest as the total number of permutations more extreme (bothhigher or lower) than the average greenscape (for a given yearor a given route) divided by the total number of permutations(n = 100 000) for each year (n = 3) or for each route (n = 32).

RESULTS

Across the entire study area, the period of spring green-upranged from 14 April to 20 June in 2013 (a drought year),from 2 April to 30 June in 2014 (heavy snowpack year) andfrom 14 February to 14 June in 2015 (a low snowpack yearwith a wet spring). Mule deer generally matched their move-ments with peaks in green-up (Fig. 2), with surfing that wasnearly twice that of simulated migrations based on the dailyrate of displacement from winter to summer range along thesame elevational gradient (Appendix S7). The vast majority(98.4%) of individual regressions between the date of peakIRG and the date of deer use were classified as surfing thatwas better than random, and 31.7% of those regressions wereconsistent with theoretically perfect surfing during the migra-tion or spring period.On average, deer received IRG values above 0.8 for fewer

days in 2013 (28.87 � 1.13, mean � SE) than in 2014(43.96 � 1.86) and 2015 (57.35 � 2.70). In 2013, long-dura-tion migrants had IRG values above 0.8 for a shorter amountof time than short- or mid-duration migrants (short =31.63 � 1.64, mid = 31.43 � 1.64 and long = 21.80 � 1.85days; mean � SE), whereas the opposite was true in 2014



(short = 39.65 � 2.57, mid = 41.38 � 3.40 and long = 49.43� 3.18 days; mean � SE) and 2015 (short = 46.33 � 5.37,mid = 60.77 � 4.41 and long = 64.06 � 3.14 days; mean �SE; Fig. 3). Days-From-Peak varied similarly across the3 years (Appendix S8).The Greenscape Hypothesis best described individual vari-

ability in green-wave surfing. Green-up duration was thestrongest predictor of IRG across individuals (DAICc of sec-ond best supported model > 6; AICc Weight = 0.92;Appendix S9). As the duration of green-up increased, IRGalso increased (b = 0.0034; 95% CI = 0.0022–0.0045; Fig. 4).Days-From-Peak was best explained by three models, contain-ing a combination of green-up duration, spring scale (recipro-cal of the green-up rate) and green-up order (cumulative AICc

weight > 0.5; Appendix S9). Days-From-Peak decreased asspring scale decreased, green-up order became more consecu-tive and green-up duration increased (Fig. 4).The greenscape was most difficult to surf in 2013

(P = 0.0013) and easiest to surf in 2014 (P < 0.0001), accord-ing to predicted IRG (Fig. 5 and Fig. S10). Despite annualfluctuations in the greenscape, 4 of 32 routes (12.5%) wereconsistently of higher quality and were predicted to result indeer receiving 9% higher average IRG than other routes. Tworoutes were consistently of lower quality and were predictedto result in deer receiving 7.5% less average IRG(Appendix S10).

DISCUSSION

Approximately a third of mule deer surfed the green wave aswell as a theoretically perfect surfer, and the vast majority ofdeer surfed the green wave better than random, providingempirical support for the GWH in a migratory ungulate.Nonetheless, surfing varied among individuals and years(Figs 3 and 4). Variation in surfing was unrelated to animalexperience or physiological state, but instead was determinedby the way the green wave progressed along the migratoryroute (i.e. the ‘greenscape’). In support of the GreenscapeHypothesis, migratory routes with longer and more

consecutive green-up allowed deer to more closely synchronisemovement with peaks in green-up. Our findings providestrong evidence that green-wave surfing plays an importantrole in the foraging strategy of mule deer, and that spatialvariability in plant phenology structures the presumed forag-ing benefits acquired during migration.In contrast to the first explicit test of the GWH in a migra-

tory ungulate wherein only a few (5.2%) animals surfed thegreen wave (i.e. red deer; Bischof et al. 2012), mule deer syn-chronised their movements during spring migration with plantgreen-up (Fig. 2). We suspect that different movement strate-gies employed by mule deer and red deer may be a functionof physiological differences between the two species that man-ifest in specialised foraging behaviours (Bell 1971; Jarman1974). Larger species such as red deer require greater forageintake and can tolerate lower quality forage, and thus may belimited more by forage quantity than smaller-bodied rumi-nants like mule deer that may be limited more by forage qual-ity (Muller et al. 2013). In addition, the distance and durationof migration varied substantially between our study (meanduration = 34.4 days; mean distance = 67 km) and that ofBischof et al. (2012; mean duration = 7.4 days; mean dis-tance = 23 km). Migration strategies of red deer and muledeer could result from different resource landscapes in Nor-way and Wyoming. Red deer in Norway might have jumpedthe green wave because parts of their migratory routes werelow-quality habitat, or had poor greenscape characteristics.Despite these differences, our work provides strong evidencethat green-wave surfing is a behaviour that is fundamental tothe foraging benefit of migration for mule deer in our studyarea.In accordance with the Adult Learning Hypothesis, we

predicted that older, more experienced individuals would bet-ter surf the green wave. Nevertheless, age had no influenceon surfing in adults. Numerous behaviours, from foragingefficiency (Gillingham & Bunnell 1989) to fawn rearing(Ozoga & Verme 1986), are learned in ungulates. Mule deerin western Wyoming have high fidelity to their migratoryroutes (Sawyer & Kauffman 2011), and the geographic

Figure 2 Date of use relative to date of peak IRG for migratory mule deer in western Wyoming, USA, in 2013 (a), 2014 (b) and 2015 (c). A perfect match

between the date of peak IRG and the date of deer use is represented with a grey 1:1 line. Individual regressions indicate that 31% of deer surfed no

different than that of a theoretically perfect surfer, whereas the vast majority of deer (98%) surfed better than random. Black points represent days during

the period of spring green-up across the study area, whereas coloured points correspond to days during the spring migration period.



location of migratory routes is thought to be learned withinthe first year of life (Nelson 1998). Therefore, learning toefficiently surf a given route might occur in the first year oflife. As we only monitored adults, we had inadequate datato examine the effect of learning in fawns or yearlings.Future work that evaluates green-wave surfing across all lifestages will provide important insights into the effects oflearning early in life, when it is often most critical (Wunderle1991; Rotics et al. 2016).A wide range of behaviours across numerous animal taxa

are affected by the nutritional or reproductive state of an indi-vidual (Martin & Lopez 1999; Heithaus et al. 2007). Nutri-tional condition has been linked to the timing of migration inmule deer (Monteith et al. 2011) and habitat selection in elk

(Long et al. 2014). Nevertheless, reproductive status andnutritional condition were unrelated to surfing. Ungulates liv-ing in seasonal environments should maximise energy intakeduring spring green-up because winter often is associated witha nutritional bottleneck (Parker et al. 2009). Therefore, it ispossible that all animals, regardless of their energetic state inlate winter, seek to maximise net energy gain via green-wavesurfing in spring to bolster somatic gain, and subsequently,reproductive success. Nevertheless, state-dependent foragingalso may occur at scales finer than our landscape-levelevaluation.In support of the Greenscape Hypothesis, animals using

routes with longer durations of green-up had more time (upto 73 days) to access heterogeneity in resources than

Figure 3 Instantaneous Rate of Green-up (IRG) experienced by mule deer in western Wyoming, USA, in 2013 (a–c), 2014 (d–f) and 2015 (g–i), by short-

(a, d, g), mid- (b, e, h) and long-duration (c, f, i) migrants. Annual IRG experienced by individual deer are shown with orange (2013), green (2014) or

purple (2015) lines; average IRG is shown in black with 95% CIs (grey lines). The light grey area represents the spring green-up period for a given year

and the dark grey area represents the mean migration period.



individuals using routes with shorter durations of green-up.Deer using routes with consecutive green-up did not have torevisit areas along their route (a behaviour that we did notobserve in our data) to track the green wave. Long green-updurations and gradual green-up rates have been linked to

improved fitness in a diverse range of ungulates (Pettorelliet al. 2005, 2007; Middleton et al. 2013; Monteith et al. 2015;Searle et al. 2015). Although we did not quantify fitness bene-fits, we hypothesise that deer migrating along profitable green-scapes will receive better forage and have higher fitness thandeer using migratory routes with poor greenscapes. Becausethe greenscape was a strong predictor of green-wave surfing,we suggest that the greenscape largely dictates the habitatquality of a given migratory route. The importance of thegreenscape highlights the value of incorporating both the spa-tial structure of resources and the temporal dynamics ofresources into the definition and quantification of habitat(Armstrong et al. 2016).In contrast to our prediction, deer deviated less from the

date of peak IRG when green-up occurred rapidly. Rapidgreen-up has been correlated with negative demographiceffects for ungulates (e.g. Middleton et al. 2013), and could beconsidered a negative attribute of the greenscape becausethere is less time when high-quality forage is available.Nonetheless, closer tracking of peaks in IRG when green-upwas rapid indicates that animals can adjust behaviourally tocompensate for rapid rates of green-up (Fig. 4). Behaviouralbuffering – in which animals adjust behaviours in response toenvironmental stochasticity – has been attributed to effectivethermoregulation (Ortega et al. 2016) and improved demo-graphic rates (Loe et al. 2016). Unlike previous studies, thebehavioural buffering of mule deer in our study was not basedupon changes in habitat use (mule deer have high fidelity to

Figure 4 Green-wave surfing of migratory mule deer in western Wyoming, USA, measured by IRG (a) and Days-From-Peak IRG (b–d) was largely

explained by characteristics of a route’s greenscape. Greenscape metrics included green-up duration (a–b), spring scale (c; the inverse of green-up rate,

interpreted as the time to reach peak green-up) and green-up order (d). Surfing metrics were calculated during the period of spring green-up in 2013–2015.An IRG value of 1 or Days-From-Peak value of 0 represent perfect surfing. Fitted values (black line) and 95% CIs (grey polygon) were calculated using

parametric bootstrapping of a univariate linear mixed-effects model that included both a random intercept and a random slope for year, allowing the effect

of the greenscape to vary by year.

(a) (b) (c)

Figure 5 Predictive maps of green-wave surfing by mule deer along their

migratory routes in 2013 (a; drought year), 2014 (b; heavy snowpack

year) and 2015 (c; low snowpack year with a wet spring) illustrate

considerable spatial and temporal variability caused by variation in the

greenscape. Map predictions depict the relationship between average IRG

and the duration of green-up along each route based upon the best fit

model across all candidate models (see Table S9a). An IRG value of 1

represents perfect surfing.



migratory routes; Fig 5), but rather based upon when theyuse habitat along their migratory route. Combining both IRG(reflecting the phenological environment and animal beha-viour) and Days-From-Peak (reflecting the behavioural com-ponent of surfing) presents great utility in unravelling howbehaviour and the environment interact to shape migration.Although the development of IRG has advanced our under-

standing of green-wave surfing and ungulate migration (e.g.Bischof et al. 2012; Rivrud et al. 2016), several analytical hur-dles remain. First, the Forage Maturation Hypothesis predictsan asymmetrical relationship between forage quality (i.e. energyintake rate) and plant biomass (Fryxell 1991), whereas IRGassumes a symmetrical relationship (see Fig. 1). Thus, IRGdoes not capture the differences in forage quality between phe-nological stages pre- and post-peak IRG. Second, because IRGis scaled (0–1), IRG cannot differentiate between habitatpatches that differ in overall plant biomass. Consequently, IRGonly quantifies timing of peak forage quality for a given loca-tion, and thus, does not necessarily reflect differences in abso-lute forage quality between locations. Refining this approach infuture studies will help improve our ability to quantify the spa-tiotemporal dynamics in forage quality for herbivores.Maps of predicted IRG depicted remarkable variability in

the greenscape across our 3-year study (Fig. 5). For instance,the drought in 2013 diminished the quality of the greenscapefor all routes, however, some routes were less affected. Never-theless, 12.5% of routes were consistently of higher qualityacross years and weather conditions. This result highlights theimportance of answering questions such as why some routesare consistently better than others, and what are the demo-graphic implications for individuals with fidelity to high- orlow-quality routes. Foraging theory predicts that animals expe-riencing higher rates of energy intake from forage will havehigher fitness than animals experiencing lower rates of energyintake (Stephens & Krebs 1986). For migratory ungulates,routes with high-quality greenscapes may lead to higher fitness,resulting in a larger proportion of the population using them.Increased use of high-quality migratory routes could manifestsimply because of maternal inheritance of routes (Nelson 1998)or strong fidelity to routes (Sawyer & Kauffman 2011). Conse-quently, future work should (1) investigate the mechanismunderlying selection of migratory routes with variable green-scapes, especially via cultural transmission, and (2) test howvariability in greenscapes underpin eco-evolutionary processessuch as density–habitat relationships, genetic structuring acrosslandscapes and vulnerability to environmental stochasticity.Our findings indicate that a migratory ungulate surfs the

green wave and in so doing, extends the amount of timeexposed to high-quality forage. Although migratory routesfunction as corridors between distinct seasonal ranges that arecritical to the persistence of migratory taxa, they also arethemselves foraging habitat that likely yield substantial nutri-tional benefits to individuals that exploit them. Researchersand conservationists must expand their view of migration tonot only consider the importance of animal movement or tem-poral resource dynamics but also how these two processesmeld together in space and time to influence fitness and popu-lation dynamics of consumers. Indeed, building upon theGWH (Drent et al. 1978) and the notion that migration

should emerge in environments that are seasonal and pre-dictable (Mueller & Fagan 2008), we demonstrate that thestructure of a resource wave (i.e. the greenscape) shapes thebenefit of migration. We also showed how annual variation inweather can affect resource waves, and in turn, the ability ofanimals to enhance resource gain via synchronised movementwith resources. Such a linkage has implications for how cli-mate change will affect the persistence of migratory taxa, andhighlights the need to incorporate temporal variability intoour conceptualisation and quantification of habitat.

ACKNOWLEDGEMENTS

Combined funding from the Wyoming Game and Fish Depart-ment, Muley Fanatic Foundation, Wyoming Governor’s BigGame License Coalition, Bowhunters of Wyoming, WyomingWildlife and Natural Resource Trust, Knobloch Family Foun-dation, Boone and Crockett Club, Wyoming Outfitters andGuides Association, US Bureau of Land Management andWyoming Animal Damage Management Board made muledeer capture/recapture and collaring efforts possible. This pro-ject was also funded by the U.S. Geological Survey WyomingLandscape Conservation Initiative and the U.S. GeologicalSurvey National Climate Change and Wildlife Science Center.The National Science Foundation Graduate Research Fellow-ship and the University of Wyoming’s Berry Fellowship pro-vide support to EOA. JAM was supported by a U.S.Department of Agriculture, National Institute of Food andAgriculture postdoctoral fellowship (grant award no. 2014-01928). We also acknowledge support from the Haub Schoolof Environment and Natural Resources, Wyoming Coopera-tive Fish and Wildlife Research Unit, Department of Zoologyand Physiology and the Program in Ecology at the Universityof Wyoming. We thank all of the volunteers who helped withdata collection related to capture and other field work. Com-ments from JR Goheen, C Riginos, TA Morrison, BR Jesmer,AC Hodge, TJ Zaffarano, AAB May, BA Oats, L Barrett, DEddy, T Avgar, G Bohrer and one anonymous reviewer helpedto improve previous drafts of the manuscript. Any use oftrade, product or firm names is for descriptive purposes onlyand does not imply endorsement by the U.S. Government.

STATEMENT OF AUTHORSHIP

EOA, MJK and KLM developed the research idea. KLM,MJK, GLF and EOA secured funding. All authors collectedthe data and SPHD managed the data. EOA and JAM anal-ysed the data. EOA wrote the first draft of the manuscriptand all co-authors contributed to revisions.

DATA ACCESSIBILITY STATEMENT

Data are available in the Dryad Digital Repository: doi:10.5061/dryad.7kc09.

REFERENCES

Albon, S.D. & Langvatn, R. (1992). Plant phenology and the benefits of

migration in a temperate ungulate. Oikos, 65, 502–513.



https://doi.org/10.5061/dryad.7kc09

https://doi.org/10.5061/dryad.7kc09

Alerstam, T., Hedenstrom, A. & Akesson, S. (2003). Long-distance

migration: evolution and determinants. Oikos, 103, 247–260.Armstrong, J.B., Takimoto, G., Schindler, D.E., Hayes, M.M. &

Kauffman, M.J. (2016). Resource waves: phenological diversity

enhances foraging opportunities for mobile consumers. Ecology, 97,

1099–1112.Avgar, T., Street, G. & Fryxell, J.M. (2013). On the adaptive benefits of

mammal migration. Can. J. Zool., 92, 481–490.Barrett, M.W., Nolan, J. & Roy, L.D. (1982). Evaluation of a hand-held

net-gun to capture large mammals. Wildl. Soc. Bull., 108–114.Bartlam-Brooks, H.L.A., Beck, P.S.A., Bohrer, G. & Harris, S. (2013). In

search of greener pastures: using satellite images to predict the effects

of environmental change on zebra migration. J. Geophys. Res-Biogeo.,

118, 1427–1437.Bell, R.H. (1971). A grazing ecosystem in the Serengeti. Sci. Am., 225,

86–93.Berdahl, A., Torney, C.J., Ioannou, C.C., Faria, J.J. & Couzin, I.D.

(2013). Emergent sensing of complex environments by mobile animal

groups. Science, 339, 574–576.Bischof, R., Loe, L.E., Meisingset, E.L., Zimmermann, B., Van Moorter,

B. & Mysterud, A. (2012). A migratory northern ungulate in the

pursuit of spring: jumping or surfing the green wave? Am. Nat., 180,

407–424.Bleich, V.C., Stephenson, T.R., Holste, N.J., Snyder, I.C., Marshal, J.P.,

McGrath, P.W. et al. (2003). Effects of tooth extraction on body

condition and reproduction of mule deer. Wildl. Soc. Bull., 233–236.Bohrer, G., Beck, P.S., Ngene, S.M., Skidmore, A.K. & Douglas-

Hamilton, I. (2014). Elephant movement closely tracks precipitation-

driven vegetation dynamics in a Kenyan forest-savanna landscape.

Mov. Ecol., 2, 2.

Bolker, B.M. (2015). Linear and generalized linear mixed models. In:

Ecological Statistics: Contemporary Theory and Application (eds Fox,

G.A., Negrete-Yankelevich, S. & Sosa, V.J.). Oxford University Press,

Oxford, United Kingdom, pp. 312–313.Boone, R.B., Thirgood, S.J. & Hopcraft, J.G.C. (2006). Serengeti

wildebeest migratory patterns modeled from rainfall and new

vegetation growth. Ecology, 87, 1987–1994.Bunnefeld, N., Borger, L., van Moorter, B., Rolandsen, C.M., Dettki, H.,

Solberg, E.J. et al. (2011). A model-driven approach to quantify

migration patterns: individual, regional and yearly differences. J. Anim.

Ecol., 80, 466–476.Burnham, K.P. & Anderson, D.R. (2002). Model Selection and

Multimodel Inference: A Practical Information-theoretic Approach, 2nd

edn.. Springer, New York.

Dingle, H. (2014). Migration: the Biology of Life on the Move, 2nd edn..

Oxford University Press.

Doherty, P.F., White, G.C. & Burnham, K.P. (2012). Comparison of

model building and selection strategies. J. Ornithol., 152, 317–323.Drent, R., Ebbinge, B. & Weijand, B. (1978). Balancing the energy

budgets of arctic-breeding geese throughout the annual cycle: a

progress report. Verhandlungen der Ornithologischen Gesellschaft in

Bayern, 23, 239–264.Fagan, W.F., Lewis, M.A., Auger-Methe, M., Avgar, T., Benhamou, S.,

Breed, G. et al. (2013). Spatial memory and animal movement. Ecol.

Lett., 16, 1316–1329.Fryxell, J.M. (1991). Forage quality and aggregation by large herbivores.

Am. Nat., 138, 478–498.Fryxell, J.M., Wilmshurst, J.F. & Sinclair, A.R.E. (2004). Predictive

models of movement by Serengeti grazers. Ecology, 85, 2429–2435.Garroutte, E.L., Hansen, A.J. & Lawrence, R.L. (2016). Using NDVI and

EVI to map spatiotemporal variation in the biomass and quality of

forage for migratory elk in the Greater Yellowstone Ecosystem. Remote

Sens., 8, 404.

Gillingham, M.P. & Bunnell, F.L. (1989). Effects of learning on food

selection and searching behavior of deer. Can. J. Zool., 67, 24–32.Grunbaum, D. (1998). Schooling as a strategy for taxis in a noisy

environment. Evol. Ecol., 12, 503–522.

Guttal, V. & Couzin, I.D. (2010). Social interactions, information use,

and the evolution of collective migration. Proc. Natl Acad. Sci. USA,

107, 16172–16177.Harrower, M. & Bloch, M. (2006). MapShaper.org: a map generalization

Web service. IEEE Comput. Graphics Appl., 26, 22–27.Heithaus, M.R., Frid, A., Wirsing, A.J., Dill, L.M., Fourqurean, J.W.,

Burkholder, D. et al. (2007). State-dependent risk-taking by green sea

turtles mediates top-down effects of tiger shark intimidation in a

marine ecosystem. J. Anim. Ecol., 76, 837–844.Hodkinson, I.D. (2005). Terrestrial insects along elevation gradients:

species and community responses to altitude. Biol. Rev., 80,

489–513.Holdo, R.M., Holt, R.D. & Fryxell, J.M. (2009). Opposing rainfall and

plant nutritional gradients best explain the wildebeest migration in the

Serengeti. Am. Nat., 173, 431–445.Hutchings, M.R., Kyriazakis, I. & Gordon, I.J. (2001). Herbivore

physiological state affects foraging trade-off decisions between nutrient

intake and parasite avoidance. Ecology, 82, 1138–1150.Jarman, P. (1974). The social organisation of antelope in relation to their

ecology. Behaviour, 48, 215–267.Johnson, D.S., London, J.M., Lea, M.A. & Durban, J.W. (2008).

Continuous-time correlated random walk model for animal telemetry

data. Ecology, 89, 1208–1215.Loe, L.E., Hansen, B.B., Stien, A., Albon, S.D., Bischof, R., Carlsson, A.

et al. (2016). Behavioral buffering of extreme weather events in a high-

Arctic herbivore. Ecosphere, 7, e01374.

Lok, E.K., Esler, D., Takekawa, J.Y., De La Cruz, S.W., Boyd, W.S.,

Nysewander, D.R. et al. (2011). Stopover habitats of spring

migrating surf scoters in Southeast Alaska. J. Wildl. Manage., 75,

92–100.Long, R.A., Bowyer, R.T., Porter, W.P., Mathewson, P., Monteith, K.L.

& Kie, J.G. (2014). Behavior and nutritional condition buffer a large-

bodied endotherm against direct and indirect effects of climate. Ecol.

Monogr., 84, 513–532.Martin, J. & Lopez, P. (1999). When to come out from a refuge: risk-

sensitive and state-dependent decisions in an alpine lizard. Behav. Ecol.,

10, 487–492.Mcnamara, J.M. & Houston, A.I. (1992). Risk-sensitive foraging – a

review of the theory. Bull. Math. Biol., 54, 355–378.Merkle, J.A., Fortin, D. & Morales, J.M. (2014). A memory-based

foraging tactic reveals an adaptive mechanism for restricted space use.

Ecol. Lett., 17, 924–931.Merkle, J.A., Sigaud, M. & Fortin, D. (2015). To follow or not? How

animals in fusion-fission societies handle conflicting information during

group decision-making. Ecol. Lett., 18, 799–806.Merkle, J.A., Monteith, K.L., Aikens, E.O., Hayes, M.M., Hersey, K.R.,

Middleton, A.D. et al. (2016). Large herbivores surf waves of green-up

during spring. Proc. R. Soc. B., 283, 20160456.

Middleton, A.D., Kauffman, M.J., McWhirter, D.E., Cook, J.G., Cook,

R.C., Nelson, A.A. et al. (2013). Animal migration amid shifting

patterns of phenology and predation: lessons from a Yellowstone elk

herd. Ecology, 94, 1245–1256.Monteith, K.L., Bleich, V.C., Stephenson, T.R., Pierce, B.M., Conner,

M.M., Klaver, R.W. et al. (2011). Timing of seasonal migration in

mule deer: effects of climate, plant phenology, and life-history

characteristics. Ecosphere, 2, Art47.

Monteith, K.L., Klaver, R.W., Hersey, K.R., Holland, A.A., Thomas,

T.P. & Kauffman, M.J. (2015). Effects of climate and plant phenology

on recruitment of moose at the southern extent of their range.

Oecologia, 178, 1137–1148.Moser, L., Fonti, P., Buntgen, U., Esper, J., Luterbacher, J., Franzen, J.

et al. (2010). Timing and duration of European larch growing

season along altitudinal gradients in the Swiss Alps. Tree Physiol., 30,

225–233.Mueller, T. & Fagan, W.F. (2008). Search and navigation in dynamic

environments - from individual behaviors to population distributions.

Oikos, 117, 654–664.



Mueller, T., O’Hara, R.B., Converse, S.J., Urbanek, R.P. & Fagan, W.F.

(2013). Social learning of migratory performance. Science, 341,

999–1002.Muller, D.W.H., Codron, D., Meloro, C., Munn, A., Schwarm, A.,

Hummel, J. et al. (2013). Assessing the Jarman-Bell Principle: scaling of

intake, digestibility, retention time and gut fill with body mass in

mammalian herbivores. Comp. Biochem. Phys. A, 164, 129–140.Mysterud, A., Langvatn, R., Yoccoz, N.G. & Stenseth, N.C. (2001). Plant

phenology, migration and geographical variation in body weight of a

large herbivore: the effect of a variable topography. J. Anim. Ecol., 70,

915–923.Nelson, M.E. (1998). Development of migratory behavior in northern

white-tailed deer. Can. J. Zool., 76, 426–432.Ortega, Z., Mencia, A. & Perez-Mellado, V. (2016). Behavioral

buffering of global warming in a cold-adapted lizard. Ecol. Evol., 6,

4582–4590.Ozoga, J.J. & Verme, L.J. (1986). Relation of maternal age to fawn-

rearing success in white-tailed deer. J. Wildl. Manage., 50, 480–486.Parker, K.L., Barboza, P.S. & Gillingham, M.P. (2009). Nutrition

integrates environmental responses of ungulates. Funct. Ecol., 23, 57–69.Pettorelli, N., Mysterud, A., Yoccoz, N.G., Langvatn, R. & Stenseth,

N.C. (2005). Importance of climatological downscaling and plant

phenology for red deer in heterogeneous landscapes. Proc. Biol. Sci.,

272, 2357–2364.Pettorelli, N., Pelletier, F., von Hardenberg, A., Festa-Bianchet, M. &

Cote, S.D. (2007). Early onset of vegetation growth vs. rapid green-up:

impacts on juvenile mountain ungulates. Ecology, 88, 381–390.Rivrud, I.M., Heurich, M., Krupczynski, P., M€uller, J. & Mysterud, A.

(2016). Green wave tracking by large herbivores: an experimental

approach. Ecology, 97, 3547–3553.Rotics, S., Kaatz, M., Resheff, Y.S., Turjeman, S.F., Zurell, D., Sapir, N.

et al. (2016). The challenges of the first migration: movement and

behaviour of juvenile vs. adult white storks with insights regarding

juvenile mortality. J. Anim. Ecol., 85, 938–947.

Sawyer, H. & Kauffman, M.J. (2011). Stopover ecology of a migratory

ungulate. J. Anim. Ecol., 80, 1078–1087.Schindler, D.E., Armstrong, J.B., Bentley, K.T., Jankowski, K., Lisi, P.J.

& Payne, L.X. (2013). Riding the crimson tide: mobile terrestrial

consumers track phenological variation in spawning of an anadromous

fish. Biol. Lett., 9, 20130048.

Searle, K.R., Rice, M.B., Anderson, C.R., Bishop, C. & Hobbs, N.T.

(2015). Asynchronous vegetation phenology enhances winter body

condition of a large mobile herbivore. Oecologia, 179, 377–391.Shaffer, S.A., Tremblay, Y., Weimerskirch, H., Scott, D., Thompson,

D.R., Sagar, P.M. et al. (2006). Migratory shearwaters integrate

oceanic resources across the Pacific Ocean in an endless summer. Proc.

Natl Acad. Sci. USA, 103, 12799–12802.Shariatinajafabadi, M., Wang, T.J., Skidmore, A.K., Toxopeus, A.G.,

Kolzsch, A., Nolet, B.A. et al. (2014). Migratory herbivorous waterfowl

track satellite-derived green wave index. PLoS ONE, 9, e108331.

Stephens, D.W. & Krebs, J.R. (1986). Foraging Theory. Princeton

University Press.

Wunderle, J.M. (1991). Age-specific foraging proficiency in birds. Curr.

Ornithol., 8, 273–324.

SUPPORTING INFORMATION

Additional Supporting Information may be found online inthe supporting information tab for this article.

Editor, Ran NathanManuscript received 4 December 2016First decision made 2 January 2017Second decision made 2 March 2017Manuscript accepted 21 March 2017



the greenscape shapes surfing of resource waves in a large migratory herbivore · 2020-04-07 ·...

Documents